Have anyone worked with O2 analyser. (urgent! its for covid patients)

[Manul]

around 5% of deviation is acceptable

Then it is really easy. If you do not care about 20.5% vs 21.0%, then UNO ADC will be enough.

EDIT: I looked up some data from Atmel and it seems that oversampling works really well with internal reference. For a slow signal, even 15-16 bits of resolution is achievable, well below 0.1mV.

[fcb]

around 5% of deviation is acceptable

Then it is really easy. If you do not care about 20.5% vs 21.0%, then UNO ADC will be enough.

EDIT: I looked up some data from Atmel and it seems that oversampling works really well with internal reference. For a slow signal, even 15-16 bits of resolution is achievable, well below 0.1mV.

Oversampling doesn't work like that.  If you've got a 10 bit ADC and you sample it 64 times, you'll end up with a 16bit result, but with 10bit resolution - divide the result by 64 and you'll still be at 10bit resolution but have averaged out some noise by virtue of the software low-noise filter.  If you add dither (noise) then you can increase resolution, but not otherwise.

Much better to scale the sensor voltage to match the ADC or use a higher resolution external ADC.  Someone else pointed out that the Atmel 328P (the proc on the UNO) has an 1.1V scale mode.

1.1V/1023=1.07mV per bit, and 1% O² change is roughly 0.65mV to 0.8mV, and if the sensor is +/-2%........whir...whir...... you'd probably get within +/-5% accuracy.

If you just use an opamp and two resistors (gain of say 14) to scale it, you should easily get with +/-2.5% overall accuracy, better still, make one of those resistors variable and you'll be able to easily calibrate the unit with simple maths and a known calibration gas (100% O²).

[Manul]

around 5% of deviation is acceptable

Then it is really easy. If you do not care about 20.5% vs 21.0%, then UNO ADC will be enough.

EDIT: I looked up some data from Atmel and it seems that oversampling works really well with internal reference. For a slow signal, even 15-16 bits of resolution is achievable, well below 0.1mV.

Oversampling doesn't work like that.  If you've got a 10 bit ADC and you sample it 64 times, you'll end up with a 16bit result, but with 10bit resolution - divide the result by 64 and you'll still be at 10bit resolution but have averaged out some noise by virtue of the software low-noise filter.  If you add dither (noise) then you can increase resolution, but not otherwise.

Much better to scale the sensor voltage to match the ADC or use a higher resolution external ADC.  Someone else pointed out that the Atmel 328P (the proc on the UNO) has an 1.1V scale mode.

1.1V/1023=1.07mV per bit, and 1% O² change is roughly 0.65mV to 0.8mV, and if the sensor is +/-2%........whir...whir...... you'd probably get within +/-5% accuracy.

If you just use an opamp and two resistors (gain of say 14) to scale it, you should easily get with +/-2.5% overall accuracy, better still, make one of those resistors variable and you'll be able to easily calibrate the unit with simple maths and a known calibration gas (100% O²).

This is not correct. Divide by number of samples is ok, but it is floating point operation, usually the decimation is done. For 64 samples, you sum them and divide by 8. And this is extra 3 bits (theoretical, but sometimes very near in practice too). What I mean about it working well with internal reference is that the internal reference + other things in the 328p ADC gives near optimal amount of noise to naturally dither from itself. A lot of ADCs react quite well to oversampling with just the internal noise, no external dither. Some better, some worse. Atmel 328p is quite good, I checked some data.

Also, yes, I wrote on the second post in this thread, that UNO (328p) has 1.1V reference.

[Adhith]

These have a finite life when you open the tin/bag they come in. So the OOM202 is quoted at >1,000,000% volume oxygen hours.  So essentially 10,000 hours (60 weeks) at 100% concentration, greater life at lower concentrations.

I was planning to use a solenoid valve to isolated the sensor from oxygen and use when needed by powering the solenoid valve. Seems like the sensor will run itself more than a year even if its exposed to 100% oxygen concentrations. So i guess I could simply skip all these solenoid valve setup right?

[Manul]

I was planning to use a solenoid valve to isolated the sensor from oxygen and use when needed by powering the solenoid valve. Seems like the sensor will run itself more than a year even if its exposed to 100% oxygen concentrations. So i guess I could simply skip all these solenoid valve setup right?

Even if it is isolated by the valve, there is still oxygen left inside the sensor from the last time valve was open, right? It could be, that there is no benefit at all. I don't know how fast the oxygen is consumed by such a sensor. I presume that it would consume some. So theoretically, if it is sealed, the concentration would drop. But probably very slowly. Not worth to think about it. They have quite good lifetime anyway.

[Adhith]

It will need to be kept away from oxygen to last, longer than the specified life, but does it matter if it only lasts a year? It thought this was just to save lives during the current emergency situation, rather than as a permanent solution. Hopefully it won't be needed for that long.

If it's needed again, then the sensor can be replaced. It has a connector on the back. If you think it might be needed in the future, you should design the machine so it can be easily replaced. There's nothing wrong with having consumable parts. You change the tyres on your car, after driving a certain distance, or the rubber perishes and this is no different.

Yeah ,you are right. on a second thought, I have made my mind to keep things simple and use the sensor full time in contact with oxygen

[Adhith]

Not necessarily. Even if it adds 5% of lifetime, is it important? I would say - don't do it. Lets think, so it is more then 1 year in 100% oxygen. Probably could survive 2 years in normal conditions. Do you really care, what will happen after 2 years? Putting solenoid just for that and possibly degrading measurment response time is definetly not worth, at least in my opinion. Protect it from high air flow, just a little to maintain exchange and it will be fine. But don't put it into one end sealed pipe...

Yes. agree with you. I'm using the sensor as it is.

[fcb]

around 5% of deviation is acceptable

Then it is really easy. If you do not care about 20.5% vs 21.0%, then UNO ADC will be enough.

EDIT: I looked up some data from Atmel and it seems that oversampling works really well with internal reference. For a slow signal, even 15-16 bits of resolution is achievable, well below 0.1mV.

Oversampling doesn't work like that.  If you've got a 10 bit ADC and you sample it 64 times, you'll end up with a 16bit result, but with 10bit resolution - divide the result by 64 and you'll still be at 10bit resolution but have averaged out some noise by virtue of the software low-noise filter.  If you add dither (noise) then you can increase resolution, but not otherwise.

Much better to scale the sensor voltage to match the ADC or use a higher resolution external ADC.  Someone else pointed out that the Atmel 328P (the proc on the UNO) has an 1.1V scale mode.

1.1V/1023=1.07mV per bit, and 1% O² change is roughly 0.65mV to 0.8mV, and if the sensor is +/-2%........whir...whir...... you'd probably get within +/-5% accuracy.

If you just use an opamp and two resistors (gain of say 14) to scale it, you should easily get with +/-2.5% overall accuracy, better still, make one of those resistors variable and you'll be able to easily calibrate the unit with simple maths and a known calibration gas (100% O²).

This is not correct. For 64 samples, you sum them and divide by 8. And this is extra 3 bits (theoretical, but sometimes very near in practice too). What I mean about it working well with internal reference is that the internal reference + other things in the 328p ADC gives near optimal amount of noise to naturally dither from itself. A lot of ADCs react quite well to oversampling with just the internal noise, no external dither. Some better, some worse. Atmel 328p is quite good, I checked some data.

Also, yes, I wrote on the second post in this thread, that UNO (328p) has 1.1V reference.

I agree that if you get some noise then you can increase the effective resolution, however this is not a guaranteed or tested behaviour - linearity/ and 'no-missing-codes' would be questionable.  I would balk at anyone suggesting you can get 15-16bit out of a 10bit converter. It might look like it if you squint hard enough, but it will suck.  Pretty much a moot point in this application as the OP is happy with +/-5% (unless the really mean +/-2.5%).

As for isolating the sensor with a solenoid valve to extend the life.  I wouldn't bother, the extra headache, complexity and delay for a marginal increase in sensor life isn't worth it, not to mention that you are pressed for time I guess??

Perhaps the OP can describe how they are using the O² percentage in their control system/loop? Or is it just to give confidence to the operator??

[Manul]

I agree that if you get some noise then you can increase the effective resolution, however this is not a guaranteed or tested behaviour - linearity/ and 'no-missing-codes' would be questionable.  I would balk at anyone suggesting you can get 15-16bit out of a 10bit converter. It might look like it if you squint hard enough, but it will suck.  Pretty much a moot point in this application as the OP is happy with +/-5% (unless the really mean +/-2.5%).

Yes, I totally agree that it is inferior to a proper high resolution ADC, but from a practical standpoint it could be a solution. And could work surprisingly well if the amount of noise happens to be optimal. For the OP application, I bet that the external noise coupling to the signal could easily be the limiting factor. I mean, it is very easy to get millivots of mains or switching noise interfering with something connected on wires. Averaging will of course solve some of that, but still. Extra resolution or op amp front end becomes irrelevant in that case.

[Adhith]

Great! Thank you all for your support. Reading from all your suggestions, I'm planning to first directly connect the sensor to the UNO and use the internal ADC function.

The thing is that, the O2 generator system itself is a pretty solid system. It could deliver more than 90% oxygen for at least a year without any attention. The sensor and controls are for safety and medical regulations approval. We are following all the guideline for the system.
We have a three led lights which shows '90-100%, 80-90%, 70-80%. So I think a bit of resolution difference is not a problem.

Since my expertise is in mechanical engg. domain and being a beginner in electronics, I need to finalize on this with you guys.

could someone help me in what to do step wise?

[Manul]

Since my expertise is in mechanical engg. domain and being a beginner in electronics, I need to finalize on this with you guys.

could someone help me in what to do step wise?

To do what? Steps for what? Anyway, as you see, quite a lot of people got involved here in the discussion, so I'm sure you will have support if some problems arise.

[Zero999]

As I said here on the forum several times enriched air is dangerous stuff.
It is non conducting and produces static electricity in each component. Even invisible ESD event can result in disaster.
There were several hospitals that were engulfed in fire due to improper installation and/or equipment.
Instead of trying to make dangerous equipment is much better to organize purchase of certified equipment.
To many this looks as opportunity to show craftmanship and ingenuity but I look at this as planting bomb nearby innocent patients.
 Just search how many people died in fires in hospital recently.

Do you really think he would be doing this if it wasn't absolutely necessary? This isn't someone's fun, hobby project, but their response to a medical emergency. You have no idea how bad the situation is in India. SARS-Cov-2 is on the rampage. More people are dying, than in developed countries because of a shortage of oxygen, hospital beds and general poor health due a large disease burden of malnutrition and diabetes.  If it was that easy to just go and buy an oxygen concentrator, I'm sure they would do it.

You're right about being cautious. Keep the room well ventilated, which should be the case anyway, as it cuts the risk of transmission of respiratory viruses and don't leave large amounts of highly flammable materials around.

Great! Thank you all for your support. Reading from all your suggestions, I'm planning to first directly connect the sensor to the UNO and use the internal ADC function.

The thing is that, the O2 generator system itself is a pretty solid system. It could deliver more than 90% oxygen for at least a year without any attention. The sensor and controls are for safety and medical regulations approval. We are following all the guideline for the system.
We have a three led lights which shows '90-100%, 80-90%, 70-80%. So I think a bit of resolution difference is not a problem.

Since my expertise is in mechanical engg. domain and being a beginner in electronics, I need to finalize on this with you guys.

could someone help me in what to do step wise?

If you do need an op-amp, there's this module which has a gain of 100.
https://www.sparkfun.com/products/9816

[Robert Smith Eco Warrior]

My brother and his work collegues made a ventilator.

It is based around a gas fuel injection system and uses the ECU to inject oxygen in a massively flexible way... Constant volume/constant pressure etc and also can do what's called CPAP (constant positive airway pressure)

I don't know if that is of any use?

I could put you in touch with them.

[daisizhou]

The theoretical effective time of the OOM2O2 sensor is about 1 year.But the actual use time is shorter,Effective measurement on ICU ventilator takes only 6 months
Because there are still interference effects of water vapor (H20) and other gases in actual use,If you use a lot, you need to replace the oxygen sensor frequently,Or buy a high-quality oxygen sensor, but the price is expensive

Therefore, OOM2O2 sensor is not the best solution,If you use a small amount temporarily,Just a temporary solution

[fcb]

If you can, you should try and involve a local electronics engineer, or group of engineers.  They will know what is available locally as well.

Whilst EEVBLOG members can design your electronics etc.., debugging it will have to be done locally - and the devil will be in the details!  You really do need to find local electronic expertise.

My understanding is that getting components/PCB's etc.. into India is difficult/slow/expensive at the best of times - so you're likely to only be able to use the parts in local subcontractors, colleges, radio stores, etc...  Out of my depth here.

[Zero999]

My brother and his work collegues made a ventilator.

It is based around a gas fuel injection system and uses the ECU to inject oxygen in a massively flexible way... Constant volume/constant pressure etc and also can do what's called CPAP (constant positive airway pressure)

I don't know if that is of any use?

I could put you in touch with them.

At the start of the pandemic, a lot of time and money was invested in ventilators, but it turned out it wasn't needed. Ventilators are only any good if clinicians are available to operate them. All the ventilators in the world are no good, without doctors and nurses.

The problem in India is lack of pure oxygen, rather than ventilators. Many COVID-19 patients just need oxygen. It is also required for anesthesia and without it, other life saving surgical procedures can't take place, causing many more excess deaths.

The theoretical effective time of the OOM2O2 sensor is about 1 year.But the actual use time is shorter,Effective measurement on ICU ventilator takes only 6 months
Because there are still interference effects of water vapor (H20) and other gases in actual use,If you use a lot, you need to replace the oxygen sensor frequently,Or buy a high-quality oxygen sensor, but the price is expensive

I repeat: does it really matter? Hopefully it won't be needed in six months time. Hopefully COVID-19 cases will have fallen by then and if not, they will have approved medical grade, oxygen concentrators.

Therefore, OOM2O2 sensor is not the best solution,If you use a small amount temporarily,Just a temporary solution

Exactly. It's not supposed to be a permanent solution, but a temporary fix, to get them out the current crisis. A life of even three months will hopefully be more than enough.

Engineering is about finding the most suitable solution to a problem, which depends on the requirements. In this case, if the OOMO2 is cheaper and more widely available, than more expensive oxygen sensors, then is the best solution to this problem.

[daisizhou]

I am a biomedical engineer,Engaged in clinical medical engineering for more than 20 years
I just combined my work experience and actual use of the effect description

When testing a sample may not find the problem,For example, 100, 1,000, 10,000, these problems will be big problems,This will directly cause the program to fail to implement and eventually fail

[Adhith]

Keeping sensor under inert gas is not recommended. This is called starvation mode and it will shorten sensor lifetime.
As I said here on the forum several times enriched air is dangerous stuff.
It is non conducting and produces static electricity in each component. Even invisible ESD event can result in disaster.
There were several hospitals that were engulfed in fire due to improper installation and/or equipment.
Instead of trying to make dangerous equipment is much better to organize purchase of certified equipment.
To many this looks as opportunity to show craftmanship and ingenuity but I look at this as planting bomb nearby innocent patients.
 Just search how many people died in fires in hospital recently.

As a personnel working in aerospace combustion, I would definitely agree with your statement. However totally disagree with your point on craftsmanship and showoff. The situation in India is so tough now, people are dying due to lack of oxygen than covid. We have the machines and fabrication stuff within our reach, so that's why we have planned to do this. Hope nothing is wrong in that considering the situation that we have.

[Zero999]

I am a biomedical engineer,Engaged in clinical medical engineering for more than 20 years
I just combined my work experience and actual use of the effect description

When testing a sample may not find the problem,For example, 100, 1,000, 10,000, these problems will be big problems,This will directly cause the program to fail to implement and eventually fail

He needs a oxygen concentrator, which is cheap and easy to build from widely available parts. It doesn't need to last for long, just a few months, until the epidemic is either over, or oxygen concentrators which meet all the relevant regulations to become available. What would you recommend?

[Adhith]

To do what? Steps for what? Anyway, as you see, quite a lot of people got involved here in the discussion, so I'm sure you will have support if some problems arise.

mainly the code for the arduino. Say, I need to turn on three different leds on the output which one ofor 95-90% rangem one for 90-95% and one for 85-90%,. So how should I setup the code?

[Adhith]

Guys, we are planning to roll out the thing next week. All stuffs are done expect the coding part. Final testing is going to happen in 2 days. Thank you all for your feed backs and suggestions.

Few people have raised some concerns over this and I just wanted to let you know, We are not doctors or expert people in this. This is just for emergency situation prevailing in the country. This is supposed to be used under the supervision of doctors till the medical oxygen demand is met by hospitals. Also, as a safety precaution for the third wave.

[Adhith]

If you do need an op-amp, there's this module which has a gain of 100.
https://www.sparkfun.com/products/9816

Great! Thank you. I will be moving to a LCD screen display eventually. So I would use this one for high accuracy.

[Manul]

Do you already have sensor? I might be able to help with some code, so do others. Have you ever done something with Arduino? Do you have people on your team with at least some coding experience?

[Adhith]

My brother and his work collegues made a ventilator.

It is based around a gas fuel injection system and uses the ECU to inject oxygen in a massively flexible way... Constant volume/constant pressure etc and also can do what's called CPAP (constant positive airway pressure)

I don't know if that is of any use?

I could put you in touch with them.

Great! best wises to him. We are not into ventilators at the moment, but definitely would keep this in mind if we need his assistance. Thank you very much

[Adhith]

quote author=Manul link=topic=282235.msg3569610#msg3569610 date=1621018086]
Do you already have sensor? I might be able to help with some code, so do others. Have you ever done something with Arduino? Do you have people on your team with at least some coding experience?
[/quote]
Yes I got the sensor a day ago. I know electronics but, not good at arduino. Have built few project around it as well but I'll take way more time to understand the terms & attributes on IDE which in this case I'm planning to accelerate.

This is my attempt on the codes. I don't kow if it works or not. Plus I dont know how I could calibrate with ambient air sensing.